pensim Package Example (Version 1.2.9)

Transcription

1 pensim Package Example (Version 1.2.9) Levi Waldron March 13, 2014 Contents 1 Introduction 1 2 Example data 2 3 Nested cross-validation Summarization and plotting Getting coefficients for model fit on all the data 4 5 Simulation of collinear high-dimensional data with survival or binary outcome Binary outcome example Survival outcome example Introduction This package acts as a wrapper to the penalized R package to add the following functionality to that package: ˆ repeated tuning on different data folds, with parallelization to take advantage of multiple processors ˆ two-dimensional tuning of the Elastic Net ˆ calculation of cross-validated risk scores through a nested cross-validation procedure, with parallelization to take advantage of multiple processors It also provides a function for simulation of collinear high-dimensional data with survival or binary response. This package was developed in support of the study by Waldron et al.[3]. This paper contains greater detail on proper application of the methods provided here. Please cite this paper when using the pensim package in your research, as well as the penalized package[2]. 1

2 2 Example data pensim provides example data from a microarray experiment investigating survival of cancer patients with lung adenocarcinomas (Beer et al., 2002)[1]. Load this data and do an initial pre-filter of genes with low IQR: > library(pensim) > data(beer.exprs) > data(beer.survival) > ##select just 100 genes to speed computation, just for the sake of example: > set.seed(1) > beer.exprs.sample <- beer.exprs[sample(1:nrow(beer.exprs), 100), ] > # > gene.quant <- apply(beer.exprs.sample, 1, quantile, probs=0.75) > dat.filt <- beer.exprs.sample[gene.quant>log2(100), ] > gene.iqr <- apply(dat.filt, 1, IQR) > dat.filt <- as.matrix(dat.filt[gene.iqr>0.5, ]) > dat.filt <- t(dat.filt) > dat.filt <- data.frame(dat.filt) > # > library(survival) > surv.obj <- Surv(beer.survival$os, beer.survival$status) Note that the expression data are in wide format, with one column per predictor (gene). It is recommended to put covariate data in a dataframe object, rather than a matrix. 3 Nested cross-validation Unbiased estimation of prediction accuracy involves two levels of cross-validation: an outer level for estimating prediction accuracy, and an inner level for model tuning. This process is simplified by the opt.nested.crossval function. It is recommended first to establish the arguments for the penalized regression by testing on the penalized package functions optl1 (for LASSO), optl2 (for Ridge) or cvl (for Elastic Net). Here we use LASSO. Setting maxlambda1=5 is not a generally recommended procedure, but is useful in this toy example to avoid converging on the null model. > library(penalized) > testfit <- optl1(response=surv.obj, + penalized=dat.filt, + fold=5, + maxlambda1=5, + positive=false, + standardize=true, + trace=false) 2

3 Now pass these arguments to opt.nested.crossval() for cross-validated calculation and assessment of risk scores, with the additional arguments: ˆ outerfold and nprocessors: number of folds for the outer level of crossvalidation, and the number of processors to use for the outer level of cross-validation (see?opt.nested.crossval) ˆ optfun and scaling: which optimization function to use (opt1d for LASSO or Ridge, opt2d for Elastic Net) - see?opt.splitval. ˆ setpen and nsim: setpen defines whether to do LASSO ( L1 ) or Ridge ( L2 ), nsim defines the number of times to repeat tuning (see?opt1d. opt2d has different required arguments.) > set.seed(1) > preds <- opt.nested.crossval(outerfold=5, nprocessors=1, #opt.nested.crossval arguments + optfun="opt1d", scaling=false, #opt.splitval arguments + setpen="l1", nsim=1, #opt1d arguments + response=surv.obj, #rest are penalized::optl1 arguments + penalized=dat.filt, + fold=5, + positive=false, + standardize=true, + trace=false) Ideally nsim would be 50, and outerfold and fold would be 10, but the values below speed computation 200x compared to these recommended values. Note that here we are using the standardize=true argument of optl1 rather than the scaling=true argument of opt.splitval. These two approaches to scaling are roughly equivalent, but the scaling approaches are not the same (scaling=true does z-score, standardize=true scales to unit central L2 norm), and results will not be identical. Also, using standardize=true scales variables but provides coeffients for the original scale, whereas using scaling=true scales variables in the training set then applies the same scales to the test set. 3.1 Summarization and plotting Cox fit on the continuous risk predictions: > coxfit.continuous <- coxph(surv.obj~preds) > summary(coxfit.continuous) Call: coxph(formula = surv.obj ~ preds) n= 86, number of events= 24 coef exp(coef) se(coef) z Pr(> z ) 3

4 preds exp(coef) exp(-coef) lower.95 upper.95 preds Concordance= (se = ) Rsquare= (max possible= ) Likelihood ratio test= 2.01 on 1 df, p= Wald test = 2.22 on 1 df, p= Score (logrank) test = 2.3 on 1 df, p= Dichotomize the cross-validated risk predictions at the median, for visualization: > preds.dichot <- preds > median(preds) Plot the ROC curve: 4 Getting coefficients for model fit on all the data Finally, we can get coefficients for the model fit on all the data, for future use. Note that nsim should ideally be greater than 1, to train the model using multiple foldings for cross-validation. The output of opt1d or opt2d will be a matrix with one row per simulation. The default behavior in opt.nested.crossval() is to take the simulation with highest cross-validated partial log likelihood (CVL), which is the recommended way to select a model from the multiple simulations. > beer.coefs <- opt1d(setpen="l1",nsim=1, + response=surv.obj, + penalized=dat.filt, + fold=5, + maxlambda1=5, + positive=false, + standardize=true, + trace=false) We can also include unpenalized covariates, if desired. Note that when keeping only one variable for a penalized or unpenalized covariate, indexing a dataframe like [1] instead of doing [, 1] preserves the variable name. With [, 1] the variable name gets converted to. > beer.coefs.unpen <- opt1d(setpen="l1", nsim=1, + response=surv.obj, + penalized=dat.filt[-1], # This is equivalent to dat.filt[,-1] + unpenalized=dat.filt[1], 4

5 > if(require(survivalroc)){ + nobs <- length(preds) + cutoff < preds.roc <- survivalroc(stime=beer.survival$os, status=beer.survival$status, + marker=preds, predict.time=cutoff, span = 0.01*nobs^(-0.20)) + plot(preds.roc$fp, preds.roc$tp, type="l", xlim=c(0, 1), ylim=c(0, 1), + xlab=paste( "FP", "\n", "AUC = ", round(preds.roc$auc, 3)), + lty=2, + ylab="tp", main="lasso predictions\n ROC curve at 12 months") + abline(0, 1) + } LASSO predictions ROC curve at 12 months TP FP AUC = Figure 1: ROC plot of cross-validated continuous risk predictions at 12 months. Note that the predictions are better if you don t randomly select 250 genes to start with! We only did this to ease the load on the CRAN checking servers. 5

6 + fold=5, + maxlambda1=5, + positive=false, + standardize=true, + trace=false) Note the non-zero first coefficient this time, due to it being unpenalized: > beer.coefs[1, 1:5] #example output with no unpenalized covariates L1 cvl U21931_at Y00285_s_at S78187_at > beer.coefs.unpen[1, 1:5] #example output with first covariate unpenalized L1 cvl U21931_at Y00285_s_at S78187_at Simulation of collinear high-dimensional data with survival or binary outcome The pensim also provides a convenient means to simulation high-dimensional expression data with (potentially censored) survival outcome or binary outcome which is dependent on specified covariates. 5.1 Binary outcome example First, generate the data. Here we simulate 20 variables. The first 15 (group a ) are uncorrelated, and have no association with outcome. The final five (group b ) have covariance of 0.8 to each other variable in that group. The response variable is associated with the first variable group b (firstonly=true) with a coefficient of 2. Binary outcomes for n s = 50 samples are simulated as a Bernoulli distribution with probability for patient s: p s = exp( βx s ) with β s,16 = 0.5 and all other β s,i equal to zero. The code for this simulation is as follows: (1) > set.seed(9) > x <- create.data(nvars=c(15, 5), 6

7 + cors=c(0, 0.8), + associations=c(0, 2), + firstonly=c(true, TRUE), + nsamples=50, + response="binary", + logisticintercept=0.5) Take a look at the simulated data: > summary(x) Length Class Mode summary 6 data.frame list associations 20 -none- numeric covariance 400 -none- numeric data 21 data.frame list > x$summary start end cors associations num firstonly a TRUE b TRUE A simple logistic model fails at variable selection in this case: > simplemodel <- glm(outcome ~., data=x$data, family=binomial) > summary(simplemodel) Call: glm(formula = outcome ~., family = binomial, data = x$data) Deviance Residuals: Min 1Q Median 3Q Max e e e e e-04 Coefficients: Estimate Std. Error z value Pr(> z ) (Intercept) a a a a a a a a a

8 a a a a a a b b b b b (Dispersion parameter for binomial family taken to be 1) Null deviance: e+01 on 49 degrees of freedom Residual deviance: e-07 on 29 degrees of freedom AIC: 42 Number of Fisher Scoring iterations: 25 But LASSO does a better job, selecting several of the collinear variables in the b group of variables which are associated with outcome: > lassofit <- opt1d(nsim=3, nprocessors=1, setpen="l1", penalized=x$data[1:20], + response=x$data[, "outcome"], trace=false, fold=10) > print(lassofit) L1 cvl (Intercept) a.1 a.2 [1,] [2,] [3,] a.3 a.4 a.5 a.6 a.7 a.8 a.9 [1,] [2,] [3,] a.10 a.11 a.12 a.13 a.14 a.15 b.1 [1,] [2,] [3,] b.2 b.3 b.4 b.5 [1,] [2,] [3,] And visualize the data as a heatmap: 8

9 > dat <- t(as.matrix(x$data[, -match("outcome", colnames(x$data))])) > heatmap(dat, ColSideColors=ifelse(x$data$outcome==0, "black", "white")) a.3 a.2 a.12 a.5 a.10 a.6 a.9 a.1 a.15 a.8 a.14 a.13 a.11 a.4 a.7 b.5 b.4 b.3 b.1 b Figure 2: Heatmap of simulated data with binary response. 9

10 5.2 Survival outcome example We simulate these data in the same way, but with response= timetoevent. Here censoring is uniform random between times 2 and 10, generating approximately 34% censoring: > set.seed(1) > x <- create.data(nvars=c(15, 5), + cors=c(0, 0.8), + associations=c(0, 0.5), + firstonly=c(true, TRUE), + nsamples=50, + censoring=c(2, 10), + response="timetoevent") How many events are censored? > sum(x$data$cens==0)/nrow(x$data) [1] 0.42 Kaplan-Meier plot of this simulated cohort: 10

11 > library(survival) > surv.obj <- Surv(x$data$time, x$data$cens) > plot(survfit(surv.obj~1), ylab="survival probability", xlab="time") Survival probability time Figure 3: Kaplan-Meier plot of survival of simulated cohort. 11

12 References [1] David G. Beer, Sharon L.R. Kardia, Chiang-Ching Huang, Thomas J. Giordano, Albert M. Levin, David E. Misek, Lin Lin, Guoan Chen, Tarek G. Gharib, Dafydd G. Thomas, Michelle L. Lizyness, Rork Kuick, Satoru Hayasaka, Jeremy M.G. Taylor, Mark D. Iannettoni, Mark B. Orringer, and Samir Hanash. Gene-expression profiles predict survival of patients with lung adenocarcinoma. Nat Med, 8(8): , [2] Jelle J Goeman. L1 penalized estimation in the cox proportional hazards model. Biometrical Journal. Biometrische Zeitschrift, 52(1):70 84, February PMID: [3] Levi Waldron, Melania Pintilie, Ming-Sound Tsao, Frances A Shepherd, Curtis Huttenhower, and Igor Jurisica. Optimized application of penalized regression methods to diverse genomic data. Bioinformatics, 27(24): , PMID: > sessioninfo() R Under development (unstable) ( r65113) Platform: x86_64-unknown-linux-gnu (64-bit) locale: [1] LC_CTYPE=en_US.UTF-8 LC_NUMERIC=C [3] LC_TIME=en_US.UTF-8 LC_COLLATE=C [5] LC_MONETARY=en_US.UTF-8 LC_MESSAGES=en_US.UTF-8 [7] LC_PAPER=en_US.UTF-8 LC_NAME=C [9] LC_ADDRESS=C LC_TELEPHONE=C [11] LC_MEASUREMENT=en_US.UTF-8 LC_IDENTIFICATION=C attached base packages: [1] splines stats graphics grdevices utils [6] datasets methods base other attached packages: [1] survivalroc_1.0.3 penalized_ survival_ [4] pensim_1.2.9 loaded via a namespace (and not attached): [1] MASS_ parallel_3.1.0 tools_